thanks|||
Probability for a rule {A,B,C} -> D is given by Support( {A,B,C,D})/Support( {A, B, C}) and is a measure of how probable the itemset {A,B,C,D} is when the items {A,B,C} is present. If you change the minimum_probability parameter, only rules which are above this threshold will be found during training. Increasing this value generates fewer rules.
Minimum_Support measures the number of cases where the itemset {A,B,C,D} appear together with respect to the total number of cases. This measures how frequent an itemset is. This is an orthogonal measure to the minimum_probability, but increasing the value will generate fewer itemsets/rules.
A good value for these parameters usually depennd on your dataset. Please let me know if you have any additional questions about these parameters.
Thanks
 
No comments:
Post a Comment